Enterprises need application performance monitoring: Charbel Khneisser, Riverbed

Charbel Khneisser, Regional Presales Director, MENA at Riverbed, addresses the challenge businesses face as performance issues plague the vast majority of IT teams rolling out eServices and digital transformation initiatives.

Question: We are a multi-division organisation that adopted an Internet facing approach and business model sometime back. Each of the divisions has built its Internet facing user interfaces and has scaled and developed their business over the years. The resources for front end development and the back end hosting applications are the same across the organisation. However, increasingly we are seeing the bounce rate of visitors going up and the engagement duration of visitors going down. We feel we are not delivering an engaging user experience on the front end and maybe we need to rebuild the end to end response channels, across the organisation. How do we get started?

Charbel Khneisser: I completely understand the challenges your business is facing as performance issues plague the vast majority of IT teams rolling out eServices and digital transformation initiatives. The main issue I believe you need to address is visibility, after all, you cannot monitor and resolve what you do not know. Performance can be impacted by a number of factors, both within the network and the application itself. So being able to pinpoint the issue is key to troubleshooting. 

A good place to start therefore, is with an end-to-end performance monitoring tool. You need to be able to isolate the domain and the best way to do this is to leverage packet capture based solutions. Packets never lie and can therefore serve as the single source of truth to help your team determine whether the problem lies in your network, front-end, application server or even the client’s endpoint devices. 

Once you have the ability to isolate the domain, you can then dive deeper into troubleshooting by leveraging specific techniques. For example, if you have identified that the challenge has resulted from network issues, you can leverage Simple Network Management Protocol based information or combine the packet-based solution that you would have already implemented with flow. On the other hand, if the bottleneck is application related, you can still use packet capture data to gain insight into what components of your web pages or web services are causing delays. 

Once you have this visibility, you then need control and optimisation. I would also recommend publishing some of your workloads on the cloud as this could help you deliver better performance at problematic locations. Here too, our acceleration solutions ensure performance is not impacted by latency. Your ultimate goal should be to move towards a hybrid model but to do this successfully, you first need to be able to benchmark performance of your users, and have a comprehensive solution for root cause identification and analysis.